Thousand Brains Project

YouTube video by Thousand Brains Project How is Monty Diffferent Than a Vision Transformer?

We’re running a Q&A series with our research team. First up:

How is Monty different from a Vision Transformer?

Here’s why learning through movement changes everything 👇

youtube.com/shorts/ff4Xi...

YouTube video by Thousand Brains Project 2025/07 - Hierarchy or Heterarchy? A Theory of Long-Range Connections for the Sensorimotor Brain

🚀 Watch @vivianeclay.bsky.social present our new paper “Hierarchy or Heterarchy?” revealing a new part of the Thousand Brains Theory describing how long‑range cortical & thalamic connections work in parallel and hierarchically to build human intelligence.

youtu.be/QIoENhFu2VU

Hands-On Intelligence: Why the Future of AI Moves Like a Curious Toddler, Not a Supercomputer “We are neural beings. Our brains take their input from the rest of our bodies. What our bodies are like and how they function in the world…

What if we've been building AI completely wrong? While tech giants burn billions training on internet data, @thousandbrains.org just created an AI that learns like a child by exploring and touching objects. It’s the real path to intelligence: gregrobison.medium.com/hands-on-int...

YouTube video by Thousand Brains Project 2025/07 - Thousand-Brains Systems: Sensorimotor Intelligence for Rapid Robust Learning and Inference

🎥 Watch @cortical-canonical.bsky.social present our new paper:

"Thousand-Brains Systems: Sensorimotor Intelligence for Rapid, Robust Learning and Inference"

Watch the full talk here: youtu.be/3d4DmnODLnE

Sensorimotor learning implemented from 20 years of neocortex research.

Thousand Brains Project Join the thousand brains community and chat with us about getting involved

10/
Thousand Brains Project is an open-source, open-research nonprofit building neocortex-based AI.

Join us on our journey.

Forum: thousandbrains.discourse.group
Roadmap: thousandbrainsproject.readme.io/docs/project...
Website: thousandbrains.org
Docs: thousandbrainsproject.readme.io

9/
Hierarchical connections are used to learn compositional models.
A new mug with a known logo means that you don’t relearn either one. Your brain composes: “mug” + “logo.” Columns at different levels but with overlapping receptive field link representations spatially.

8/
So what’s the thalamus doing?
It’s not just a relay. It helps convert sensory input from body-centered to object-centered coordinates. Essential for modeling the world via reference frames.

7/
Neocortical columns don’t think in egocentric space, “left of hand.” They think in object-centric space, “on the handle.” The thalamus helps do this translation. This is a key new proposal about the role of the thalamus.

6/
The Thousand Brains Theory already outlined several key proposals:
Each column builds its own model.
They vote.
They form consensus.
No need to wait for a final decision at the top.

5/
This paper argues that the neocortex is not a strict hierarchy. It has many non-hierarchical connections, each of which serves an essential role in modeling and interacting with the world.

4/
Our new view recasts cortical hierarchy as composition, not feature extraction. Hierarchy is used to combine known parts into new wholes.

an incomplete view of how the neocortex uses hierarchy

3/
Classically, the neocortex is viewed as a hierarchy: low levels detect edges, high levels recognize objects. But that model is incomplete. There are several other important connections most people overlook. But they are crucial for sensorimotor intelligence.

YouTube video by Thousand Brains Project 2025/07 - Hierarchy or Heterarchy? A Theory of Long-Range Connections for the Sensorimotor Brain

2/
TLDR; Watch @viviane give a presentation about this paper here: youtu.be/QIoENhFu2VU
Or read the plain language explainer here: thousandbrains.medium.com/hierarchy-or...

1/
🚨Another New Paper Drop! 🚨 “Hierarchy or Heterarchy? A Theory of Long-Range Connections for the Sensorimotor Brain”

👇 Dive into the full thread 🧵
arxiv.org/abs/2507.05888

YouTube video by Artem Kirsanov A Fundamental Unit Of Intelligence

🔥 Want to understand how the neocortex builds intelligence?

Artem Kirsanov made a great video on the Thousand Brains Theory, the foundation of everything we’re building at here Thousand Brains Project!

🎥 youtu.be/Dykkubb-Qus
#Neuroscience #AI #ThousandBrains #Neocortex

Thousand Brains Project Join the thousand brains community and chat with us about getting involved

/16
Thousand Brains Project is an open-source, open-research nonprofit building a new type of machine intelligence based on principles of the neocortex.

Join us:

Forum: thousandbrains.discourse.group
Roadmap:
Docs: thousandbrainsproject.readme.io thousandbrainsproject.readme.io/docs/project...

GitHub - thousandbrainsproject/tbp.tbs_sensorimotor_intelligence: Configs and code to replicate experiments from our paper, "Thousand Brains Systems: Sensorimotor Intelligence for Rapid, Robust Learni... Configs and code to replicate experiments from our paper, "Thousand Brains Systems: Sensorimotor Intelligence for Rapid, Robust Learning and Inference" - thousandbrainsproject/tbp.tbs_sen...

/15
💾 Open source
• Replicate our results: github.com/thousandbrai...
• Code & docs: github.com/thousandbrai...
Fork it, break it, improve it.

/14
Voting across modules
Multiple learning modules share hypotheses; consensus arrives >2× faster without losing accuracy. Imagine two eyes, two fingers, or a sensor grid collaborating in real time.

/13
🏃‍♀️ Movement matters
A simple curvature-following policy + a hypothesis-testing policy cut inference steps ~3× vs. random walks. Monty can leverage its learned models to perform principled movements to resolve uncertainty.

/12
Robust inference and generalization
Monty recognized all 77 YCB objects and their pose with 90+% accuracy, even with noise, novel rotations. Even showing the object in a never-before-seen color doesn’t phase Monty.

/11
Learning modules can quickly reach consensus through voting about their most likely object and pose hypotheses, rather than having to integrate over time with multiple sensations.

/10
Motor commands from all the learning modules tell the system where it should observe next, causing a subsequent movement to observe that location.

/9
Learning Modules: a semi-independent modelling system that builds models of objects by integrating sensed observations with object-relative coordinates derived from the body-centric sensor locations.

/8
This new architecture comprises of the following subsystems that communicate using the cortical messaging protocol (CMP)
Sensor Modules: that observe a small patch of the world and send it to the learning module.

/7
This is a new type of machine learning architecture based on principles derived from 20+ years of research into the neocortex.

/6
Few-shot learning
After just 8 views per object Monty hits ~90 % accuracy.

/5
Monty learns continually with virtually no loss of accuracy as new objects are added to the model. Deep learning famously suffers from catastrophic forgetting in those settings.

/4
Green AI is here: Monty slashes compute needs in training and inference, yet beats a pretrained, finetuned ViT on object and pose tasks. It used 33,888× fewer FLOPs than ViT trained from scratch and 527,000,000× fewer than a pretrained ViT.

3/
A remarkable result: Monty achieves the same accuracy as a Vision Transformer while using 527 million times less computation, and it does so without suffering from catastrophic forgetting.

YouTube video by Thousand Brains Project 2025/07 - Thousand-Brains Systems: Sensorimotor Intelligence for Rapid Robust Learning and Inference

/2
TLDR; Watch @cortical-canonical.bsky.social give a presentation about this paper here: youtu.be/3d4DmnODLnE

Or read the plain language explainer here: thousandbrains.medium.com/thousand-bra...